New Feature

• [HIVEMALL-171] - Tracing functionality for prediction of DecisionTrees
• [HIVEMALL-231] - Create Subarray UDF
• [HIVEMALL-248] - UDFs for Kuromoji stoptags
• [HIVEMALL-253] - UDF map_roulette() created
• [HIVEMALL-261] - Support argmax/argmin UDF
• [HIVEMALL-262] - Support argsort/argrank UDF
• [HIVEMALL-276] - [Umbrella] official support for xgboost v0.90
• [HIVEMALL-281] - Support max_by, min_by, majority_vote UDAF

Bug

• [HIVEMALL-165] - array_remove UDF throws exception when the first argument is null
• [HIVEMALL-232] - [DOC] Fix typo in the Top-K document
• [HIVEMALL-235] - Fix [Double|Int|Float|Long]ArrayList bug in expansion where size=0
• [HIVEMALL-236] - to_json cause KryoException/NullPointerException with ArrayList due to Kryo bug
• [HIVEMALL-238] - from_json UDF does not work for top-level map object
• [HIVEMALL-259] - [BUG] feature_binning does not work properly under certain conditions
• [HIVEMALL-260] - Remove dependencies to Scala library in xgboost classifier
• [HIVEMALL-268] - train_ffm(features,label) causes NPE while train_ffm(features,label,'') works fine
• [HIVEMALL-274] - Wrong target variable name in the step-by-step tutorial

Improvement

• [HIVEMALL-43] - [MIXSERV][Umbrella] ProtocolBuffer-based MixMessage serialization
• [HIVEMALL-107] - Move `spark-shell` into Docker
• [HIVEMALL-121] - Add `-libsvm` formatting option for `feature_hashing`
• [HIVEMALL-178] - NaN/missing value/null values handling in RandomForest
• [HIVEMALL-200] - Add an option to return PartOfSpeech in tokenize_ja and tokenize_cn
• [HIVEMALL-226] - Move hivemall.fm and hivemall.mf packages to under hivemall.factorization
• [HIVEMALL-230] - Revise Optimizer Implementation
• [HIVEMALL-233] - RandomForest regressor accepts sparse vector input
• [HIVEMALL-234] - Fix mismatched default value between EtaEstimator and UDF option description
• [HIVEMALL-245] - Refactor RandomForest for Sparse Data handling
• [HIVEMALL-246] - Add validation for feature UDF
• [HIVEMALL-249] - Fix fmeasure UDAF to support any integers
• [HIVEMALL-251] - Add option to return PartOfSpeech information for tokenize_ja
• [HIVEMALL-258] - Add UDF to convert feature/label in Libsvm format
• [HIVEMALL-271] - Make xgboost hyperparameters configurable
• [HIVEMALL-272] - Refine xgboost_predict implementation
• [HIVEMALL-273] - Support xgboost v0.90
• [HIVEMALL-275] - Fix xgboost module to create DMatrix from CSRMatrix
• [HIVEMALL-278] - Bump up matrix4j dependencies to v0.9.1
• [HIVEMALL-279] - Support xgboost v0.90 hyperparameters

Task

• [HIVEMALL-25] - [DOC] Update UserGuide for v0.6
• [HIVEMALL-153] - [Umbrella] Release Hivemall v0.6.0
• [HIVEMALL-214] - Update userguide for General Classifier/Regressor example
• [HIVEMALL-237] - Add usage in ML function reference page
• [HIVEMALL-241] - Change minimum Java version requirements from 7 to 8
• [HIVEMALL-242] - Drop support for Spark 2.1
• [HIVEMALL-243] - Fix nominal variable handling in DecisionTree and RegressionTree
• [HIVEMALL-244] - Support Java9, Java11(LTS), and Java12
• [HIVEMALL-247] - Recommend hive.optimize.cte.materialize.threshold=2 in Hive tuning tips
• [HIVEMALL-267] - Drop Spark Dataframe support
• [HIVEMALL-269] - Use matrix4j for the matrix module
• [HIVEMALL-270] - Remove rand_gid/rand_gid2 macro
• [HIVEMALL-283] - Bump up netty version to 4.1.42.Final

Sub-task

• [HIVEMALL-26] - Add documentation about Hivemall on Apache Spark
• [HIVEMALL-27] - Add documentation about Xgboost support in v0.6.0
• [HIVEMALL-56] - Add documentation about Similarity/Distance functions
• [HIVEMALL-158] - Refine deprecated userguide contents
• [HIVEMALL-159] - Add documentation about One-hot encoding
• [HIVEMALL-250] - Add documentation about binarize_label UDTF